INDEX

Explanations

mentions of specific names and job titles

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

DragonMagazine

-0.79

 robbers

-0.78

 thieves

-0.62

BOOK

-0.61

tumblr

-0.60

problem

-0.60

align

-0.60

 capitals

-0.60

soDeliveryDate

-0.59

 womb

-0.59

POSITIVE LOGITS

McK

0.89

 Bennett

0.87

McC

0.84

 Nelson

0.84

 Vand

0.83

 Doyle

0.83

 Alexander

0.83

 Walters

0.83

 Michel

0.83

 Stewart

0.83

Activations Density 1.882%

mentions of specific names and job titles

No Comments

No Known Activations